See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/251722242

The politics of the program: MS Word as the invisible

grammarian

Article in Computers and Composition · December 2002

DOI: 10.1016/S8755-4615(02)00142-1

CITATIONS READS

38 604

2 authors, including:

Timothy McGee
Rider University
4 PUBLICATIONS 69 CITATIONS

SEE PROFILE

All content following this page was uploaded by Timothy McGee on 10 June 2020.

The user has requested enhancement of the downloaded file.

 Computers and Composition 19 (2002) 453–470

The politics of the program: ms word as

the invisible grammarian夽
Tim McGee a,∗ , Patricia Ericsson b
a
Philadelphia University, School House Lane and Henry Avenue, Philadelphia, PA 19144, USA
b
Michigan Technological University, Houghton, MI 49931, USA

Abstract
Because of its widespread availability to writing students, the grammar and style checker in ms word
deserves a thorough critique. Although recent scholarship has addressed general issues surrounding
grammar and style checkers, we investigate this particular program in depth, focusing on its theoretical
underpinnings. We contrast the approaches to grammar and style embodied in the software with those
found in current composition pedagogy and conclude with suggestions that go beyond customizing
the Grammar Checker to advocating more thorough discussions of the very notions of stylistic and
grammatical correctness.
© 2002 Elsevier Science Inc. All rights reserved.

Keywords: Composition; Correctness; Current traditional rhetoric; Grammar; Language; Linguistics; Natural
language processing; Pedagogy; Style

The most profound technologies are those that disappear. They weave themselves into the fabric
of everyday life until they are indistinguishable from it.
Mark Weiser (1991)

1. Introduction

The idea for this article was conceived accidentally. As we exchanged drafts for a confer-
ence presentation, we were entertained by some of the inane recommendations that our word

夽
Tim McGee and Patricia Ericsson are veteran techno-rhetoricians who have collaborated on several confer-
ence presentations and co-authored an article published in Kairos: A Journal for Teachers of Writing in Webbed
Environments that is based on their experience as co-directors of the Dakota State University Online Writing Lab.
∗
Corresponding author.
Email addresses: mcgeet@phila.edu (T. McGee), plericss@mtu.edu (P. Ericsson).

8755-4615/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved.
PII: S 8 7 5 5 - 4 6 1 5 ( 0 2 ) 0 0 1 4 2 - 1
454 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470

processor’s grammar and style checker suggested. Soon, casual entertainment led to questions:
Who programs these checkers? What approach to language does this automated system rep-
resent? What might this ever-present corrective force be teaching students? Does a grammar
and style checker support or undermine what composition teachers are trying to accomplish?
As our list of questions grew longer, we realized that we had, serendipitously, found a topic
for another conference presentation and for this article.
As our discussions progressed, Cynthia L. Selfe and Richard Selfe’s (1994) enjoinder from
“The Politics of the Interface” echoed regularly: Composition teachers need to be more than
users of technology; they need to “think carefully about the implications of its use within
their own classrooms” (p. 496). The Selfes encouraged us “to teach students and ourselves
to recognize computer interfaces as noninnocent physical borders. . . cultural borders. . . and
linguistic borders” (p. 495).
Joel Haefner’s (1999) “Politics of the Code” recommended an initial focus. Haefner quoted
Theodor Nelson’s claim that “a computer language is a system for casting spell,” and concluded
his article by asserting that if Nelson is right, “then English instructors in computer-supported
classrooms need to know something about the context and the necromancers of the code”
(p. 338).
Because we have concerns about the uses of technology, particularly word processing soft-
ware in the composition classroom, we focused on the style and grammar checking tool in ms
word, the most widely used word processing software in the world. Its ubiquity is unquestion-
able. And its Grammar Checker (MSGC)1 is now the product of an increasingly sophisticated
branch of Microsoft’s research efforts, the Natural Language Processing Group, whose partic-
ular linguistic approach exerts an undeniable, though largely invisible, influence upon writers.
The specifics of that approach and the philosophies of language that it embodies begged some
investigation.
We also became intrigued by how machine-checking of grammar and style alters the very
concept of teaching grammar.2 Its default status (“Check grammar as you type”) makes gram-
mar a primary concern by foregrounding correctness even while writers are in the drafting
stage. It gives grammar and style (both narrowly defined by this digital program) a prece-
dence, at least temporally, over content, mode of discourse, or other rhetorical concerns.3 This
precedence could well have a negative effect on the development of writing ability. Although
the field of composition and rhetoric enjoys some diversity of opinion as to what constitutes
“best practice,” there is considerable agreement about some classroom activities that do not
improve student writing, first among which would have to be grammar instruction (Braddock,
Lloyd-Jones, & Schoer, 1963; Hartwell, 1985).4 The stronger indictments assert that formal
grammar instruction is not just ineffective, but positively harmful if it consumes limited in-
structional time that could be dedicated to activities that actually help students become better
writers. Where, however, does an artificially intelligent checker fit into that understanding of
the place of grammar instruction in the composition curriculum? Does the checker provide in-
struction? Take up time? A related question is when, within a writing process, is the best time
for a student to attend to those issues to which Grammar Checkers do attend? Mike Rose (1984)
listed editing “too early in the composing process” as a primary cause of writer’s block.5 And,
significantly, among the five distinct meanings of grammar that Patrick Hartwell identified, to
which does the MSGC attend?
 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470 455

Finally, we considered the strategies that users and teachers might employ to maximize
the help (or minimize the harm) that the Grammar Checker in word could actually cause
in (or around) a composition classroom. Various authors have recognized the opportunities
for customizing the programs we use (Haefner, 1999; Haist, 1999; Selfe & Selfe, 1994).
Caroline Haist concluded her thorough analysis of Grammar Checker in word 97 with a list
of recommendations, the first of which is to “have students proofread their work to eliminate
as many errors as possible before using the Grammar Checker,” a wise suggestion given her
finding that “the more errors there are in a sentence, the more likely that Grammar Checker
is to provide incorrect suggestions” (“Recommendations” section, 1999, par. 2). However,
because the default setting in recent versions of word is “Check grammar as you type,” most
users will do just the opposite of what Haist recommends. Furthermore, as Haefner noted,
“customization assumes a single user for the software, not a lab or classroom environment
where many students use the same software copy, the same workstation, many times a day”
(p. 334). Finally, there is a limit to how much customizing any program will accept. As Fred
Kemp (1992) wrote, “Computer software. . . no matter what flexibility it may claim or what
ability to accept ‘user definition’ or modifying parameters, can never escape the instructional
attitudes and even the ideology of its programmers and designers” (p. 9–10).
Our argument, then, is that the Grammar Checker in ms word represents an especially
important piece of software for composition teachers to use critically for several reasons: its
ubiquity, its near invisibility, its increasing power, its theoretical mismatch, and, in most cases,
its actual conflict with and possible undermining of pedagogies that are now considered most
effective for improving student writing.

2. Ubiquity of Microsoft’s Grammar Checker

In a purely numbers game, Microsoft wins the software war hands down. In the early 1990s,
Microsoft claimed to have 35 million users of its operating system and software; however, since
the monopoly controversy of the late 1990s, specific numbers are difficult, if not impossible, to
find. Industry experts estimate Microsoft’s current share of the word processing software market
at somewhere between 80 and 90%,6 and in 1997 Microsoft was aiming to become a major force
in the educational software market, reversing their position earlier in that decade (Microsoft
makes, 1997). The ever-expanding “Microsoft in Education” branch of its web site illustrates
the corporation’s desire to promote its presence in the educational market. Simply by virtue of
its ubiquity, Microsoft gets more “teachable moments” than the English teachers do. There are
now just over 500,000 English teachers in the United States. Because word is on millions of
desktops and Grammar Checker is turned on by default, it has many more, practically invisible,
“over your shoulder”7 opportunities to be a grammar teacher than the typical English teacher—
formerly one of the only purveyors of grammar instruction in a student’s writing experience.

3. Whose grammar is it?

This change in instructional personnel is no small matter. It reminded us of the Selfes’

warning that unless “those who are familiar with language and learning theory, who understand
456 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470

issues raised by technology studies and cultural studies” pay close attention to the design of
programs, “interface design will continue to be dominated primarily by computer scientists and
will lack perspectives that could be contributed by humanist scholars” (1994, p. 498). However,
the issue here goes deeper than interface design to become not only a matter of who is teaching
grammar and how they are teaching it, but a matter of what they mean by “grammar” in the
first place.
Some users may assert that the Grammar Checker in word is nothing but a mechanical
delivery system for the grammar found in handbooks, and, therefore, represents no cause for
alarm. Others, however, while accepting the view of MSGC as a mechanical handbook, would
consider that to be an indictable offense, as handbook grammar is sometimes a dumb-downed
version of “real” grammar,8 and simply tossing it at a struggling writer is far from effective
pedagogy. We, however, contend that the content of MSGC is, in fact, quite different from
a traditional handbook. And although it might be comforting if we could demonstrate that
the grammar those computational linguists poured into MSGC represents an impoverished or
debased form of what English teachers (wise humanists, they) have been dispensing for years,
the reality is not so simple.
Yes, some MSGC content comes straight out of handbooks, and is therefore every bit as
beneficial or useless as what can be found in some “school grammars.” And some points
of careful usage that better grammar handbooks try to preserve get flattened by the binary
Grammar Checker, thereby depriving users of a subtle marker of formal tone. For example,
several recent handbooks insist upon “which” for nonrestrictive relative clauses and recommend
“that” for restrictive clauses, but will allow “which” for restrictive clauses, noting that such
usage indicates a more informal tone. The MSGC, however, takes a strictly binary approach. It
flags any “which” that is not bracketed by commas, thereby suggesting that restrictive clauses
ought to use a “that.” It is difficult to determine whether the MSGC does this because it’s
easier for the program to handle a simple binary or because the computational linguists had
loaded in the more restrictive prescriptions of Strunk and White, rather than the more accurate
description of more recent handbooks.
Nevertheless, the Grammar Checker in word embodies a far more sophisticated
approach to grammar than most users have noticed (sophisticated in terms of computational
linguistics (CL), yet increasingly mongrelized in terms of grammars). Paradoxically, how-
ever, its increased sophistication may make it more dangerous than ever before. And al-
though it may seem ironic if the wise humanists were suddenly to be trumped by those
with less enlightened views of language, the history of English grammar would simply be
repeating itself. Writing about the rise of prescriptive grammar in the eighteenth century,
Albert C. Baugh and Thomas Cable (1993) reported that Joseph Priestley’s linguistic
“tolerance, and good sense” lost out in the marketplace to Robert Lowth who was “much
more conservative in his stand, a typical representative of the normative and prescrip-
tive school of grammarians” (p. 269). About the flurry of grammars published in the late
eighteenth century, Baugh and Cable remarked that “most of these books were the work
of men with no special qualification for the thing they attempted to do” (p. 270). Two
centuries later, a similar event may be occurring as those who are highly qualified to
do CL start assuming functions that had traditionally been the bailiwick of English
teachers.
 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470 457

4. Critical use

Heeding Haefner’s call to examine “the context and necromancers of the code” behind
the Grammar Checker requires looking into Microsoft’s Research Group, first established
in 1991. The establishment of these research labs indicated Microsoft’s interest in “speech
recognition, futuristic user interfaces and 3-D graphics-cutting-edge technologies” (Buderi,
1999, p. 45). The branch of Microsoft’s Research Group that produced the MSGC is the
Natural Language Processing Group, consisting of about 50 researchers—almost all of whom
are computational linguists. Although some consider the field of CL synonymous with Natural
Language Processing (NLP), the editors of The Handbook of Natural Language Processing
assert “a clear demarcation between domain-specific theory on the one hand and practical
development of computational language processing systems on the other.” It is the latter,
purely utilitarian approach that they put at the center of NLP, calling it the “least ambitious”
of those fields that link computational ideas and human cognitive functions, limiting itself
largely to producing “practical tools for the design of input–output modules” (Dale, Moisl, &
Somers, 2000, pp. vi–vii). An epitome of such practicality would be a tool to check that aspect
of written discourse that so many writers have some insecurity about—their grammar.
The Natural Language Processing Group hit paydirt when they produced a grammar and
style checker better than the one that Microsoft had been licensing from Inso.9 When Mi-
crosoft installed its own checker in word 97, they not only improved their product and saved
some money, but, according to Dan Ling (former Research Director at Microsoft and now
the Vice-President of Microsoft Research), the sentence parsing program now built into every
copy of word was “the foundation for building lots of different natural language tools” (qtd. in
Buderi, 1999, p. 47). This foundation has much to do with Microsoft’s long range goals regard-
ing usability and office automation, but very little to do with the teaching of writing. George
E. Heidorn, a researcher in Microsoft’s NLP group, described the “product orientation” in the
NLP research group as an “enormous benefit.” Instead of being interested only in theoretical
ideas and “interesting linguistic examples,” researchers now pay attention to “real text as writ-
ten by real people” (Heidorn, 2000, p. 205) due to the requirements of such a practical project.
Their attention, however, is still that of linguists, and the linguist’s interest in language is far
different from that of the rhetorician and the composition teacher,10 and even farther still from
that of the typical student writer. As Eric Johnson (1997) concluded, some “checkers have the
ability to use sophisticated means to identify some grammatical blunders. . . but because they
sometimes propose embarrassingly bad changes, the user must exercise great care in accepting
suggestions” (par. 8). The “typical student writer” often has neither the grammatical expertise
nor the patience to take the “great care” Johnson deemed necessary. In a similar vein, Caroline
Haist found “most of the explanations Grammar Checker provides. . . helpful,” but judged that
“some are confusing and, in at least a couple of cases, just plain wrong” (1999, Introduction
section, par. 6).
To become critical users of a program designed not by teachers of writing but by computa-
tional linguists, and made not for students of writing, but for workers in Microsoft’s generic
“Office,” we must look closely at the approaches to language used by the NLP Group and then
consider the points of contact and conflict with the approaches to language used by composition
teachers.
458 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470

5. Code upon code

Haefner warned that “how programmers go about writing code structures the way we (can)
use computers” and Structured Programming’s “grounding in the American corporate environ-
ment [has] profound implications for the kinds of writing students do in computer-supported
classrooms and the ways we teach writing” (1999, p. 337). The tenor of those implications are
captured in Haefner’s earlier assertion that “the cultural imperative catalyzing SP has always
been corporate productivity and profitability” (p. 330). Sociolinguist Lesley Milroy (1998)
reminded us that English prescriptive grammar results from the fact that “by the eighteenth
century Britain needed a standardized language to meet the needs of geographically scattered
colonial government servants and to facilitate mass education” (p. 96) and that “since the goal
of codification is to define a particular form as standard, this process entailed intolerance of
a range of choices which speakers and writers had hitherto taken for granted” (p. 95). So, a
computerized Grammar Checker, even one that was well designed with student writers in mind,
would be likely to include two different kinds of codification constraining the choices of the
novice writer.11 And, as we will show, the Grammar Checker in word, because of its peculiar
history, is anything but a program well designed for student writers.

6. More real than ideal

Building on Seymour Papert’s idea that computers are “naturally heuristic,” Fred Kemp
claimed that we cannot “computerize any activity without having to (1) completely rethink
the activity, including all the assumed behaviors that have become virtually invisible, and (2)
discover in the new perspectives afforded by such rethinking, possibilities for actions that were
never possible through the old perspectives” (qtd. in Rickly, 2001, p. 10). Unfortunately, the
NLP Group has not followed this “naturally heuristic” path in programming what users get in
the MSGC. The real goal of the NLP Group is natural language processing: a user being able
to employ natural language (actual speech) to interface with a computer. The implementation
of the MSGC was just a happy coincidence for the Group.
To be responsible scholars and teachers, however, we must take note that no matter how
theoretically sound and linguistically appropriate an approach the Group takes to natural lan-
guage processing, its currently marketable byproduct (the MSGC) does not take a theoretically
sound, rhetorically appropriate, pedagogically suitable approach to an ideal grammar and style
checker. As Eric Johnson (1992) concluded:
The ideal grammar and style checker will be produced by a team consisting of linguists,
computer programmers, and writing teachers. Linguists are obviously needed to develop the
algorithms of word identification and sentence parsing. Programmers create the computer code
to implement the algorithms. No less important are the writing teachers. They know the kind
of advice that is likely to produce better writing. (par. 19)

A team composed of the members that Johnson suggested, aided by rhetoricians and com-
positionists, a team setting out to computerize an ideal grammar and style checker would, no
doubt, aspire to what Papert and Kemp proposed: They would “rethink the activity, [investigate]
 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470 459

assumed behaviors. . . discover new perspectives,” and come up with an approach that truly
would give them “a new lens through which to view the whole picture,” a lens that would show
them “new ways of conceiving the very mission itself and supporting it” (qtd. in Rickly, 2001,
p. 10).
It is unfortunate for all users that the current MGSC is not “ideal” or even close to it. There
is no “new lens,” no “new ways of conceiving” grammar or style in the MSGC. Although the
programming might be complicated and the interface slick, users get recycled, often confusing
advice about grammar and a mixed bag of style suggestions that don’t take into account
current thinking on the grammar itself, good rhetorical theory, or pedagogical considerations.
As we illustrate further, the problems of the MGSC are numerous, often dangerous, and largely
invisible.

7. More power than intelligence

The Grammar Checker in ms word 97 and word 2000 (MSGC) is, indeed, a very different
beast from the checkers found in previous versions of Microsoft’s word processing software.
Although the casual user may have noticed little change, one reviewer claimed that word 97 was
“the first word processor available that comes with a Grammar Checker worth using” (Lowe,
1997, p. 36). The new checker was considerably more sophisticated than its predecessors, at
least in terms of CL—a kind of sophistication, however, that may only exacerbate the problem
caused by the software’s rhetorical naivete (a naivete that is understandable given that the
Grammar Checker was not designed by composition teachers for student writers, and that the
ultimate goal of its designers was not Standard Written English, but speech recognition).
According to Heidorn, “the NLP system that is behind the Microsoft Grammar Checker
[but not yet fully implemented] is a full-fledged natural language processing system that
is also intended to be used for many other applications. It consists of a programming lan-
guage and a runtime environment that are both specifically tailored to the needs of an NLP
system” (2000, p. 182). One part of that processing system is described as a “bottom-up,
multipath, chart-parsing algorithm [that] also makes use of both probabilistic information and
a heuristic-scoring scheme to guide the process to a quicker and better termination point”
(p. 184).
Although such power and sophistication have caused scholars to grant that the MSGC and
other recent checkers demonstrate “undeniable gains in functionality over their predecessors,”
critics still concluded that “their reliability and accuracy. . . have improved only minimally”
(Vernon, 2000, p. 331).
An example of increased functionality with questionable reliability can be seen in the Gram-
mar Checker’s newfound ability to revise—not just flag—passive constructions. For example,
when MSGC flags “Bill was left by Mary,” it suggests revising it to “Mary left Bill.” Unfor-
tunately, however, when the current checker encounters the sentence “Bill was left by the side
of the road,” it suggests “The side of the road left Bill.”
Due to gaffes like that,12 many remain unimpressed with how the program handles language
analysis, despite recognizing that the ability to revise passive sentences into active ones is quite a
feat of artificial intelligence. Also, few users suspect that revising passive to active constructions
460 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470

Fig. 1. Diagram of NLP analysis components. Source: Microsoft Research: Current Research at <http://research.
microsoft.com/nlp/analysis.asp>.

is just the tip of an iceberg of potential functions hidden within Microsoft’s Natural Language
Analysis System, only a fraction of which are enabled in the MSGC currently on the market.
Figure 1 depicts all six of the stages that Microsoft’s full-fledged NL analysis system can
perform upon text in their research lab.
At the top of the diagram three different kinds of texts are depicted: the input text to be
analyzed, a dictionary (with input and output arrows occurring twice), and a shelf of books,
the large knowledge base they call “MindNet” (also showing both input and output arrows).
Below the texts are the six stages of analysis, representing the progression from lexical, through
syntactic, then semantic, and culminating in discourse analysis.
The MSGC in currently available versions of word is limited largely to lexical and syntactic
processing, whereas the NLP system behind the checker is well equipped to handle some kinds
of semantic processing as well. Semantic processing occurs in the third stage of the Microsoft
NLP analysis system, the syntactic portrait. In that stage, the NL analysis system attempts to
“produce more reasonable attachments for some modifiers.” Heidorn explained how semantic
reattachment “makes use of semantic relations that are produced automatically by analyzing
the text of definitions and example sentences in an online dictionary. These semantic relations
are stored in a rather large knowledge base that is now known as MindNet.” That is how
 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470 461

Microsoft NLP system knows “that it is more likely that a telescope is an instrument for seeing
a bird, rather than a part of the bird in the sentence ‘I saw a bird with a telescope”’ (p. 185).13
However, such a complex system cannot yet be packed in its entirety into the commercial
word processor. “Fortunately,” according to Heidorn, “some applications of the Microsoft NLP
system, such as the current Grammar Checker, can do an adequate job without doing semantic
reattachment and, therefore, can be spared the added cost of that processing” (2000, p. 185)—
adequate, perhaps, for the mature writer who wants help spotting the occasional slip in grammar
or usage, but far from adequate for students with a tenuous grasp on Standard Written English.
If the system behind the MSGC is, in fact, so linguistically sophisticated, why would such a
system mistake “the side of the road” for the unnamed agent who had abandoned Bill? This is
partly due to gaps in the system’s collection of “factoids” and “multiword entries” (MWEs),
bits of information that the currently installed system does possess about certain individual
words or word groups. Thanks to a factoid, the very program that suggests revising “Bill was
left by Mary” to “Mary left Bill” does not suggest that “Bill was left by April” be revised
to “April left Bill.” Why? Because the software knows that “April” can represent a month.
However, “by the side of” would not be a factoid but a multiword preposition stored as its own
“MWE” in the dictionary. That is how MSGC knows to suggest “behind” as a replacement
for “in back of.” Apparently, however, the system does not yet possess “by the side of” as an
MWE for “beside,” because it never suggests that revision.
In addition to the increased powers of syntactic analysis, another improvement, first added
in word 2000, is the software’s ability to flag successive nouns in groups of more than three.
Listed under the “Style” options, this feature allows the user to guard against “Strings of several
nouns that may be unclear, as in ‘The income tax office business practices remained the same”’
(Microsoft, 2000). This feature is one of five new checking options that Alex Vernon said
are “likely inspired by wordperfect” (2000, p. 331). Whatever the inspiration, the added
proscription against noun + noun + noun represents a new level of style (or grammar)14
checking of the sort addressed by Joseph Williams, who warns against the long compound
noun phrase as a source of ambiguity (Williams, 2000, p. 91). This new addition to the MSGC
could represent a slight tonic to its overall rhetorical naivete because linguistic expertise of
the sort that Hartwell describes as “developed by psycholinguistic studies of comprehension”
appears to be finding its way into the Grammar Checker. This is the sort of perspective that
William J. Vande Kopple (1998) felt linguists could offer to compositionists because “research
in linguistics and discourse analysis can provide powerful generalizations about how readers
will respond to characteristics of sentences and texts,” generalizations about which “students
deserve to know” (p. 6). So some of the added functionality in word 2000 appears to be getting
rhetorically smart. Again, however, just how well suited this sort of instruction is for student
writers is debatable. Vernon noted that “the target market for these products [is] business and
the adult professional,” (p. 331) a remark similar to the response some instructors have had
toward Joseph Williams’s Style: Ten Lessons in Clarity and Grace; namely, it represents a
style manual particularly well suited to the mature writer of the impenetrable prose sometimes
known as bureaucratese or sociologese, but it is a style manual not particularly well suited
to the novice writer whose difficulties have more to do with invention and fluency than with
turgidity. So much for the visible influences in the most recent version of the MSGC. We now
turn to some of its unseen influences.
462 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470

8. Disappearing powers of MSGC

Two ways that the MSGC exerts its influence are the corporate cache of Microsoft itself and
the interface of word—powers that are all the more influential as they have become unnoticed
norms. Although tarnished by the monopoly trials that began in the late 1990s, the corporate
image of Microsoft still shines as the standard bearer of the operating system and software
world. Industry watchers keep a close eye on Microsoft’s every move, its stock serving as a
bellwether for the entire technology sector. Its brand name alone gives the Microsoft Grammar
Checker a kind of influence that is difficult (perhaps impossible) to calculate.
word’s interface is another visible factor with almost invisible power, as its polished sur-
face has become so familiar that it has disappeared below the threshold of our notice. Granted,
occasional new additions catch some attention, not all of which is positive. (Why the animated
paper clip seems to arouse such antipathy in some users is hard to say; is it the simple intru-
siveness or the bizarre coquetry of its eye movements?) In any event, the overall slickness of
word’s interface is something now so familiar we no longer notice it as an incredibly polished
product constantly persuading us of its quality and dependability.
Other aspects of the program’s influence are encoded into its default settings. Because so
many users remain unaware of the options available in the Grammar Checker, millions have
grown as accustomed to the squiggly green lines as they had become to the red ones signaling
spelling errors. Grammar and style are constantly being checked in real time except for those
users savvy enough to have turned off that option. And even the savvy user gets grammar and
style advice automatically when choosing to spellcheck a document.
Some might argue that the green squiggles do render the checker visible, especially for those
accustomed to word processing before the arrival of word 97. Many users may simply learn to
ignore the green line or, more accurately, think that they have learned to ignore the green line.
However, not acting on the line’s recommendation or not bothering to inquire into its meaning
is not the same as being uninfluenced by it. Some users have reported switching “which” to
“that” whenever a “which” is flagged, but learning to ignore the full-sentence underlines (as
they usually indicate “long-sentence,” and those users don’t mind writing long ones).
Although some users may have doubts about artificial intelligence after seeing the Spiel-
berg/Kubrick film, AI, the intelligence built into the MSGC isn’t nearly as ominous as that
in the movie, but does lend another invisible force to the program. The invisible intelligence
of a supposedly smart machine has a definite appeal. Fred Kemp commented that to many,
the “computer. . . projects the universal ethos of science itself” (1992, p. 10). When this ethos
and a trust in artificial intelligence are combined with the supposition that the smart people at
Microsoft know more about grammar and style than the typical user, the Grammar Checker
gains considerable power. More significant, however, may be the issue of consistency. Un-
like teachers who may occasionally contradict themselves and who frequently contradict each
other (making “Can I use ‘I’?” one of the most frequently asked questions in every composi-
tion class), MSGC is as consistent as only a machine can be. In every situation, the Grammar
Checker will respond identically to the same grammar and style issues. Students, moreso
than the general population, have a near desperate need for certainties and “right” answers;
a computer program gives them those certainties more readily than all-too-human English
teachers.
 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470 463

Of particular interest to us and to others who are interested in the politics of the interface,
code, and software is the entire politics of standardized language. This issue is invisible to
most users, especially to student writers. Given that their school-based language instruction is
largely prescriptive and that few, if any, precollege students have been let in on the secret that the
prescriptions of Standard English are neither natural, nor logical, nor productive of language
that is “better” (in any meaningful sense of the word) than nonstandard Englishes, most simply
accept the authority of this smart machine to police their grammar, just as they have heretofore
accepted the authority of their smart teachers. Although many might smart from this power
relationship, for most, it has operated invisibly, as just one of the many givens of the school
environment. For them, the situation simply could not be otherwise. It is just what is. “Good
grammar” is a fact of school life, and the MSGC is licensed to check their grammar by the
same invisible authority that licenses the rest of the adult world they are being socialized into.
Although all previous issues are vitally important to us as composition teachers and scholars,
perhaps none is as significant as the possible misfit the MSGC might have with our objectives
as teachers of writing. Not surprisingly, our conclusions on that question are as discouraging
as the majority of our findings.

9. The MSGC contradicts good pedagogy

Throughout the recent history of composition (as well as the entire history of rhetoric),
various approaches to teaching writing have existed simultaneously. No one theory or pedagogy
has ever taken hold as “gospel.” This variety, according to James Berlin (1996), tends to
proliferate “in proportion to the freedom tolerated in the society involved” (p. 138). Despite
this multiplicity, we can look at current composition theory and pedagogy to get a sense of
what is now considered best practice—an important notion to consider if we are going to judge
whether the MSGC might support or undermine those attempting to enact such practice.
Best practice is characterized by a concern for the social nature of writing and an eye to the
total ecology in which writing is incubated, produced, and consumed. Current best practice
acknowledges that writing—whether it be invention, process, production or consumption—is
a multilayered, collaborative, often digitally enacted, postmodern ecology. Marilyn Cooper
(1989), one of the first scholars to posit an ecological model of writing, explained that this
kind of a system “postulates dynamic interlocking systems that structure the social activity of
writing” (p. 7). These ecological systems are continually in flux, steadily being created and
recreated by readers, writers, and teachers in a complex web of interacting possibilities that
those involved encounter. No matter which subtle flavor it takes on, best practice in the new
millennium is decidedly social.
The nearly polar opposite of a social/ecological approach to teaching composition is the
pedagogy commonly known as current traditional rhetoric. This method of teaching is char-
acterized by an overwhelming concern with forms and obsessive attention to precise, correct
language. Although scholars as notable as Robert Connors (1997) objected to the term current
traditional (pp. 4–7), Debra Hawhee (1999) claimed that current traditional rhetoric forms “a
discipline in both senses of the word” because current traditional rhetoric has the ingredients
of a discipline: “a subject, a ‘body of knowledge,’ a body of precepts for students to learn
464 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470

and follow”; in addition, it disciplines students by “constructing them as aberrant individuals”

(p. 521). The discipline of current traditional rhetoric is passed down from teacher to student
and then is replicated in the students who become the next generation of teachers. The precepts
of current traditional rhetoric are encoded in handbooks and workbooks that were considered
ineffective for the teaching of writing as early as the 1880s (Connors, 1997, p. 117).
Contemporary critics of Grammar Checkers have commented that “style and grammar pack-
ages are generally based on an overly narrow—and erroneous—vision of ‘correct’ language
use” (Selfe & Selfe, 1994, p. 489) or have claimed that simply tossing a handbook at a strug-
gling writer represents some pretty bad pedagogy. Kemp posited that “computers can process
text in only the most superficial of senses; computers cannot grasp the meaning in the text”
(1992, p. 14).15 The least effective, empirically dismissed approach to the improvement of
writing is the approach presented by the MSGC. It is concerned primarily with prescriptive
issues of usage and surface concerns of style. Even in its screen appearance, it harkens back
to the red pencil of the obsessive English teacher who bled over “mistakes” and paid little or
no attention to the quality of thinking. The Grammar Checker in word 97, according to one
reviewer, “attempts to be a full-fledged writing coach” (Campbell, 1997, p. 109), but the bulk
of its comments are based on a digital, right/wrong, binary paradigm that make writing seem
to be exactly what it is not: “Writing is not a mistake to be corrected, something broken to
be fixed, a gap to be filled, or a wrong to be righted” (Kemp, 1992, p. 23). Sadly, MSGC is
primarily a current traditional machine—a machine that looks dangerously smart, especially
to users insecure about grammar and usage.
Even though materials on the Microsoft web site tout word 2000 as having “new features
[that] also make it an excellent tool for collaboration” (Word 2000, 2001, par. 1) and explain
how word can be used to enhance the writing process, making word a useful teaching tool
requires that both students and teachers overlook the most obvious critique that word makes—
red and green highlighted commentary on correctness. In addition, the ability to change styles
may seem like a computer-smart guarantee that the writer is going to “get it right” for her
particular audience simply by setting a level of formality that the computer verifies. This
surface consideration of language that is audience appropriate is almost laughably simplistic.

10. Disarming the Microsoft necromancer

At the end of our efforts to “pay attention” to this particular software, we are tempted to
encourage all writing teachers to throw away their copies of ms word and to ban it from
their schools. However, if we actually had the power to instigate such a revolution, the most
prevalent grammar and style checker on our computer desktops would already be different.
Despite this inability, we believe composition teachers are not powerless, even in the face of
the Microsoft giant; we believe we can play David to this Goliath. Slaying the giant probably
isn’t possible (and perhaps now that we know so much more about the software, it might not
even be desirable), so how can composition teachers minimize the harms it could cause?
Carolyn Haist provided several suggestions to minimize the harm of the MSGC at the end
of her 1999 article, and others have provided similar suggestions (Spinks, Wells, & Meche,
1997; Vernon, 2000). In giving our presentation on this topic at the Computers and Writing
 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470 465

Conference in 2000, we still felt quite confident in providing a list of things that a composition
teacher could do to ameliorate problems MSGC might present. And we still believe that a list of
steps can help a composition teacher—to some degree. However, because of what our critique
has uncovered, we find using practical, how-to, or step-by-step guides too insignificant—rather
like just throwing rocks at Goliath instead of aiming a slingshot at him.
What can we provide as a metaphorical slingshot? What can we employ to make our aim
more accurate and our impacts more substantive? We believe the answer lies in knowledge.
To begin with, leaving decisions about grammar up to Microsoft is simply unacceptable. All
English teachers need more than a basic understanding of prescriptive grammar; we need a
rich understanding, informed by scholarship like that of Patrick Hartwell (1985) and others
who have considered grammar in a more complex way. We need to understand that “grammar”
can be used politically to limit access to position, power, and wealth. We need to understand
the subtleties of grammar far better than most of us do.
We need to spend time digging around in software before using it. When a new grammar-and-
style checker appears, we need to take time to dig into it, rummage around in its “options”16
and “defaults” to see what kind of a beast it is. (Most of us need to do this immediately with the
checkers that are currently on our desktops and in our labs—even if they aren’t new.) We need
to be confident enough with the technology to “play” with it, to open it up for inspection, and
to think about what we see and what it means for us and for students in our classes. Mindlessly
accepting a piece of software is irresponsible—even if everyone in the world is using it, even
if we can’t really change it, even if we’re afraid of breaking it. We are in complete accord with
what Selfe told us in her 1998 CCCC address: “As composition teachers, deciding whether
or not to use technology in our classes is simply not the point—we have to pay attention to
technology” (Selfe, 1999, p. 415, original emphasis).
In this same address, Selfe also told us that we need to get students involved in critiques of
technology:
Composition teachers, language arts teachers, and other literacy specialists need to recognize
that the relevance of technology in the English studies disciplines is not simply a matter of
helping student work effectively with communication software and hardware, but, rather, also
a matter of helping them to understand and be able to assess—to pay attention to—the social,
economic, and pedagogical implications of new communication technologies and technological
initiatives that affect their lives. (p. 432)

This might appear to be difficult, but we believe that it doesn’t have to be. If teachers are
confident that the software can’t really be broken (and it can’t—Microsoft has made sure of
that), then we can feel free to ask students to help us investigate it. Whatever age students we
teach, many know more about playing with computers than we do, and they are usually more
eager to do it. We can ask them to look at the options that the MSGC offers. Once they are
inside the software, we can encourage questions about the various grammar, usage, and style
choices that Microsoft offers. We think that more than a few students will notice that under
the “Formal” option in word 97, the checker has all but three possibilities turned on—and one
possibility that is not on is to check for gender-specific words. If we ask them to check into
the explanations of the grammar and style options, they can check to see if their notions of
“Commonly confused words” match up with the ones Microsoft provides.17
466 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470

The opportunities that this kind of investigation can provide us are practically unlimited. It
gives us a chance not only to talk about grammar in context, but also to think about and talk
with students about the politics of Microsoft, how the checker was constructed, and what it
might mean for different kinds of students from differing backgrounds.
Finally, those of us in positions to influence teacher education programs must insist that the
courses in these programs include up-to-date composition theory, broad-based and complex
instruction in grammar, investigations of the ideological underpinnings of technology, and
enough hands-on work with computers and software that teachers are eager to go below the
primary interface. Those of us in positions to influence hiring practices need to make sure that
the people we hire can interrogate technology.
In the end what we need is quite revolutionary, even though the revolution might not include
slaying Goliath. Our revolution must go to the heart of how we think about and interact with
technology; this revolution is about knowledge, education, and action. We need a revolt in the
ranks—people currently teaching with these technologies need to take a critical interest. And
we need to include our students in this uprising.
But none of this is really new—those in-the-know have been calling for these kinds of
changes for at least a decade. What is new, we believe, is our call for these changes following
a thorough critique of a piece of software that can powerfully affect our teaching. We are
confident that our critique, one that uncovers a largely invisible influence in our teaching lives,
will spur some to action. We hope that what we have uncovered makes the invisible grammarian
of MSGC less mysterious, less able to cast spells. We hope we have wrested some power from
the Microsoft necromancers and put it back into more capable hands—yours.

Notes
1. Although its full and most accurate name might be Microsoft’s Grammar and Style
Checker, for the sake of simplicity we will refer to the tool that appears in ms word 97
and word 2000 as the Grammar Checker or MSGC for short.
2. In addition to altering the concept of teaching grammar, machine-checking of grammar
could, to some extent, alter the concept of writing itself. As Anne Herrington and Charles
Moran (2001) concluded their critique of the machines now being used to score student
essays, “Writing to the machine. . . creates what is for now an unprecedented and un-
natural rhetorical situation. . . . Writing to the machine desensitizes us as writers” (pp.
496–97).
3. Proponents of the Grammar Checker might argue that because users can first choose one
of five “styles,” the Grammar Checker actually gives priority to the notions of audience
and occasion; however, many users remain unaware that there are five styles from which to
choose or that each of those can be customized—few of the participants in two workshops
we have conducted even knew about these options.
4. Hartwell admitted that the grammar issue was still controversial in 1985 and listed papers
defending the teaching of formal grammar as well as those attacking it. A 1995 Com-
position Chronicle article with the title “Grammar Making a Comeback in Composition
Teaching” (McCleary, 1995), attests to grammar’s perseverance. The controversy con-
tinues to this day, but, thanks in part to Hartwell’s work, the complex notion of grammar
 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470 467

has been unpacked to the point of including five distinctly different meanings, a few of
which we will employ later in this article.
5. In his attempt to identify the “cognitive variables involved in writer’s block,” Rose (1984)
listed six reasons that some writers block, four of which could well be exacerbated by
using MSGC in its default mode. In addition to editing too early, three other causes are
these: “(1) the rules by which they guide their composing processes are rigid, inappro-
priately invoked, or incorrect. . . ; (5) they invoke conflicting rules, assumptions, plans,
and strategies; and (6) they evaluate their writing with inappropriate criteria or criteria
that are inadequately understood” (p. 4).
6. Estimates vary widely depending on how the data are sampled; for example, Microsoft’s
dominance appears smaller when law offices figure heavily in the sample as Word
Perfect still commands a large share of the legal word processing market, holding 82%
of that market as recently as 1998 (Phelps, 1998, par. 2.).
7. An online Product Enhancements Guide for word once proudly asserted that word can
“act as the reader over your shoulder as you type.” That document, previously available at
<http://www.microsoft.com/TechNet/Word/prodfact/wd97peg.asp>, has been removed.
8. Hartwell (1985), expanding on “The Three Meanings of Grammar” that W. Nelson Fran-
cis offered in 1954, provided five distinct meanings, including grammar 1, “The Grammar
in Our Heads,” and grammar 2, “The Scientific Grammar of Linguists.” The third, which
Francis referred to as “Linguistic Etiquette,” according to Hartwell, “is, of course, not
grammar at all, but usage” (p. 110). Grammar 4 is school grammar, a user friendly but
slightly inaccurate version of scientific grammar that teachers use with students, is found
in most handbooks. Hartwell described the “rules” of “common school grammars” as
“inadequate to the facts of written language” (1985, p. 119).
9. That product had first been available for PCs in 1989 as CorrectText Grammar
Correction System from Houghton–Mifflin, later from Inso Corporation. No slouch of
an application, it was one of two commercial systems identified by Robert Dale (1996)
as “the first products to use anything related to the parsing technologies developed in
the research field.” He went on to predict, accurately, that “as machines become more
powerful, and as broad-coverage grammars become more feasible, we can expect to see
more of the CPU-hungry techniques developed in research labs finding their way into
products” (section 7.5.1, par. 4).
10. The gulf between linguists and compositionists is perhaps greatest around the grammar
issue. Discussing what he calls the “romantic” position that “stylistic grammars. . . have
little place in the teaching of composition,” Hartwell (1985) asserted that “this position
rests on a theory of language ultimately philosophical rather than linguistic” and sug-
gested that we “witness, for example, the contempt for linguists in Ann Berthoff’s The
Making of Meaning: Metaphors, Models, and Maxims for Writing Teachers” (p. 124).
William J. Vande Kopple (1998) argued that “In the last twenty or fifty years, research
on language has gone from an area that specialists in composition and rhetoric took quite
seriously to one that specialists now pay little attention to” (p. 4).
11. Some will argue that constraint, especially in the area of standard English and preferred
usage, is exactly what novice writers need; however, most compositionists will agree that
such constraint is best applied late in the composing process, not “as you type.”
468 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470

12. The Grammar Checker also duly flags “Me and my friends went to the store,” but suggests
that the writer should have written “My friends and me went to the store.”
13. This “rather large knowledge base” represents the solution to the persistent shortcoming
of the artificial intelligence use of computers that Kemp (1992) explained as follows:
“Despite more than 30 years of effort, artificial intelligence has foundered on the problem
of implanting the kinds of experience into computers which provides a linguistic context
capable of handling the extraordinary contextualism of natural language” (p. 16).
14. Hartwell’s grammar 5 is “Stylistic Grammar,” which he says enjoys “two fully articulated
positions” that he labels the “romantic” and the “classic,” offering Joseph Williams’s
Style: Ten Lessons in Clarity and Grace as an example of the latter position.
15. Interestingly, Kemp (1992) attributed the ultimate failure of “artificial intelligence use
of the computer” to the fact that “the problem lay in the computer’s inability to employ
natural language” (p. 13). The MSGC represents the best commercial effort, so far, to
address that part of the problem, but, as its gaffes indicate, its ability to employ (and
analyze) natural language is still quite limited.
16. Getting into the innards of the MGSC is as easy as clicking on the “Options” tab when the
grammar and spelling checker is running. From there, clicking on “Settings” will open
a range of possible choices. To see even more of the workings, clicking on “Help” and
typing “grammar and style options” in the search field will give the user a moderately
informative explanation of each possible grammar and style option.
17. We are still amused and amazed that “augur/auger” appears in this list of commonly
confused words but that “affect/effect” does not. (Perhaps the computational linguists
developed their list in collaboration with a school of Mining and Divination.) We assume
students will also experience some incredulity.

Tim McGee directs the Instructional Technology Program at Philadelphia University. His
scholarly interests include computer-based writing pedagogy, the rhetoric of science, and
argument from the locus of quantity. He has published in Text Technology, JTW: The
Journal of Teaching Writing and the section on oral presentations for the Allyn and Bacon
Guide to Writing.

Patricia Ericsson is a PhD candidate in rhetoric and technical communication at Michi-

gan Technological University. Her scholarly work focuses on agency in technological
environments, composition theory, rhetoric, and assessment. She has published in Text
Technology and ACE Journal, and she has also contributed chapters to various books.

References

Baugh, Albert C., & Cable, Thomas. (1993). A history of the English language (4th ed.). Englewood Cliffs: Prentice
Hall.
 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470 469

Berlin, James. (1996). Rhetorics, poetics and cultures: Refiguring college English studies. Urbana, IL: National
Council of Teachers of English.
Braddock, Richard, Lloyd-Jones, Richard, & Schoer, Lowell. (1963). Research in written composition. Urbana, IL:
National Council of Teachers of English.
Buderi, Robert. (1999). Software’s ultimate sandbox. Technology Review, 102(1), 44–51.
Campbell, George. (1997). Word 97. PC World, 15(2), 109–110.
Connors, Robert. (1997). Composition-rhetoric: Backgrounds, theory, pedagogy. Pittsburgh: University of
Pittsburgh Press.
Cooper, Marilyn. (1989). The ecology of writing. In Marilyn Cooper & Michael Holzman (Eds.), Writing as social
action (pp. 1–13). Portsmouth, NJ: Boynton/Cook.
Dale, Robert. (1996). Computer assistance in text creation and editing. In Ronald A. Cole, Joseph Mariani, Hans
Uszkoreit, Annie Zaenen, & Victor Zue (Eds.), Survey of the state of the art in human language technology.
Retrieved March 21, 2001, from <http://cslu.cse.ogi.edu/HLTsurvey/ch7node7.html#SECTION75>.
Dale, Robert, Moist, Herman, & Somers, H. L. (2000). Handbook of natural language processing. New york: Marcel
Dekker.
Haefner, Joel. (1999). The politics of the code. Computers and Composition, 16, 325–339.
Haist, Caroline. (1999, November). An evaluation of Microsoft Word 97’s Grammar Checker. Syntax in the schools,
16(2). Retrieved July 1, 2001, from <http://www2.pct.edu/courses/evavra/ATEG/Mono/Haist/GCHECK.htm>.
Hartwell, Patrick. (1985). Grammar, grammars, and the teaching of grammar. College English, 47, 105–127.
Hawhee, Debra. (1999). Composition history and the Harbrace College Handbook. College Composition and
Communication, 50(3), 504–523.
Heidorn, George E. (2000). Intelligent writing assistance. In Robert Dale, Hermann Moisl, & Harold Sommers
(Eds.), Handbook of natural language processing (pp. 181–207). New York: Marcel Dekker.
Herrington, Anne, & Moran, Charles. (2001). What happens when machines read our students writing? College
English, 63, 480–499.
Johnson, Eric. (1992). The ideal grammar and style checker. Text Technology, 2(4). Retrieved July 1, 2001, from
<http://www.dsu.edu/∼johnsone/ideal.html>.
Johnson, Eric. (1997). The current state of grammar and style checkers. Text Technology, 7(1). Retrieved July 1,
2001, from <http://www.dsu.edu/∼johnsone/grammar.html>.
Kemp, Fred. (1992). Who programmed this? Examining the instructional attitudes of writing-support software.
Computers and Composition, 10, 9–24.
Lowe, Doug. (1997). More Word 97 for Windows for dummies. Foster City, CA: IDG Books.
McCleary, Bill. (1995). Grammar making a comeback in composition teaching. Composition Chronicle: Newsletter
for Writing Teachers, 8(6), 1–4.
Microsoft. (2000). Microsoft Word 2000 [Computer Software] online help files. Redmond, WA: Microsoft
Corporation.
Microsoft makes move into educational software. (1997, September 24). Lubbock-Avalanche Journal. Retrieved
July 1, 2001, from <http://lubbockonline.com/news/092597/microsof.htm>.
Milroy, Lesley. (1998). Myth 12: Bad grammar is slovenly. In Laurie Bauer & Peter Trudgill (Eds.), Language
myths. London: Penguin Books.
Phelps, Alan. (1998). Why not Word? Competitors still challenge Microsoft’s word processor dominance. In
Smart computing in plain English. Retrieved July 12, 2001, from <http://www.smartcomputing.com/editorial/
article>.
Rickly, Rebecca. (2001, May). Technology, institutional assessment, and big brother. Council of Writing Program
Administrators President’s Newsletter, 8–10.
Rose, Mike. (1984). Writer’s block: The cognitive dimension. Carbondale, IL: Southern Illinois University Press.
Selfe, Cynthia L. (1999). Technology and literacy: A story about the perils of not paying attention. College
Composition and Communication, 50(3), 411–436.
Selfe, Cynthia L., & Selfe, Richard, Jr. (1994). The politics of the interface: Power and its exercise in electronic
contact zones. College Composition and Communication, 45(4), 480–504.
Spinks, Nelda H., Wells, Barron W., & Meche, Melanie. (1997, October). Customizing grammar checking. Business
Education Forum, 24–26.
 470 T. Mcgee, P. Ericsson / Computers and Composition 19 (2002) 453–470

Vande Kopple, William J. (1998, April 1–4). Current research on language and its status with composition teachers.
Paper presented at the annual meeting of the conference on college composition and communication (ERIC
ED418408).
Vernon, Alex. (2000). Computerized Grammar Checkers 2000: Capabilities, limitations, and pedagogical
possibilities. Computers and Composition, 17, 329–350.
Weiser, Mark. (1991, September). The computer for the twenty-first century. Scientific American, 94–104.
Williams, Joseph. (2000). Style: Ten lessons in clarity and grace (6th ed.). New York: Longman.
Word 2000. (2001). Microsoft in education. Retrieved July 1, 2001, from <http://www.microsoft.com/
education/tutorial/classroom/o2k/word.asp>.

View publication stats